In this episode of Legally Bond, Kim speaks with Bond cybersecurity and data privacy attorney Jessica Copeland. Jessica, who is chair of Bond's AI Committee, discusses AI governance and how the use of generative AI tools will impact the legal field.
To register for the AI Summits mentioned in this episode, click on the links below:
NYC, March 5
Long Island, March 6
In this episode of Legally Bond, Kim speaks with Bond cybersecurity and data privacy attorney Jessica Copeland. Jessica, who is chair of Bond's AI Committee, discusses AI governance and how the use of generative AI tools will impact the legal field.
To register for the AI Summits mentioned in this episode, click on the links below:
NYC, March 5
Long Island, March 6
Hello and welcome to Legally Bond, a podcast presented by the law firm Bond Shannigan King. I'm your host, Kim Wolf Price. Today we're excited to welcome back Jessica Copeland, a member of the firm from the Buffalo office, who chairs the firm's Cyber Security and Data Privacy Practice Group and is a litigator and part of our IP practice group. Most relevant for today is that Jessica also chairs the firm's Artificial Intelligence Committee, a group working to develop firm policies and procedures around this rapidly evolving technology, and she's doing work in this area as well for clients. So welcome back to the podcast, Jessica. So good to have you with us.
Speaker 1:Thank you, Kim. Thanks for inviting me back. Anytime I love having you as a guest. So when you've been on the podcast in the past, we've spent a lot of our time discussing cybersecurity and data privacy. I then feel like I need to change all of my passwords immediately after we have the conversation. Those are two critical areas for business to devote resources to, and now I'm sure that will, of course, come up. But do you mind if we focus a good part of our conversation on another thing business should pay attention to, and that's artificial intelligence?
Speaker 2:Absolutely, and it's become a priority for most organizations, cross-sector, so it makes sense to talk about it here.
Speaker 1:All right. So, since it's actually been some time since we've had you back for a full episode, would you mind giving listeners a reminder of your background? And that's whatever you'd like to talk about hometown, family, education, whatever you'd like to discuss, Sure.
Speaker 2:So I grew up in Long Island and went to college at NYU where I obtained a mathematics degree, which is highly relevant in the conversation of artificial intelligence. Is is built on large language models and algorithms, which I hold near and dear to my heart. I went to law school at St John's in Queens, new York, and I started practicing at a boutique patent litigation firm, morgan and Finnegan in New York City, fast forward, where I met my husband. I moved to Buffalo, where Chris is closely a native, in that he grew up outside of Buffalo but attended UB Law School and we started a family here.
Speaker 2:My career took a different trajectory and I started to work mostly in commercial litigation with a focus on intellectual property, but I missed sinking my teeth into technology, and it was around the time I would say 2011, 2012, where I started to focus to self-train on data privacy and cybersecurity matters for clients and organizations. I obtained a certificate in cybersecurity leadership through Carnegie Mellon and then my career path took me to Bond, where I started in 2019. And I'm the chair of the cybersecurity practice group here and I still keep a healthy docket of patent litigation.
Speaker 1:Fantastic. I love the math major in you talking about algorithms and how fun that all is, which is fantastic. I also know you're a loyal sports fan, and so I'll just say here's to next football season, even though the Super Bowl hasn't happened for us whatever. But I want to know when we're going to a Knicks game, because they're doing all right, and maybe a Liberty game down the road.
Speaker 2:Yeah, Well, I mean for the Knicks fans, as, again, I grew up in Long Island going to the Godin. So go New York, go, and yes, you did hear my Long Island accent. Many people say where did it go? It's here. I just mask it mostly.
Speaker 1:Yes, exactly. Well, I'm ready to go to the Godin with you anytime. Let's go see. Go Knicks, go. Well, thanks for adding that. I really enjoy this piece of the podcast where we get to learn about our guests, and I think that the listeners do as well. So, thank you. So now to the more serious topic of artificial intelligence. It seems every attorney publication and listserv that I receive this is now a constant topic, but I guess maybe we should back up Right? So what is artificial intelligence? Do you want to give us some background?
Speaker 2:Sure. So artificial intelligence in a general manner is really a sort of computer learned ability to function like the human brain. Machine learning is a similar fashion. It processes datasets and produces outcomes or recommendations. So think about how long we have all been streaming Netflix or listening to music on Spotify. And isn't it a miracle that Netflix knows that I'm going to like legally blonde who because I watched Blueless, believe it or not, and it's connected the dots there. And similarly, when I listen to Spotify, I have recommendations of the similar genre music. So if I play to Peshmoan, it will then shuffle to Erasure or something similar to that. So, and yes, that is a wide spectrum of interest.
Speaker 1:Which I love, and we may send each other Taylor Swift and Brandi Carlisle songs too. So it's all over the place. It all works. It all works.
Speaker 2:But on the artificial intelligence topic, and really why everyone is discussing this new technology, is that we are now talking about generative artificial intelligence, and it goes one step further, and that's the large language model that not only just predicts an outcome, but does it in the fashion of communication or imagery. So it's not just the text that we're prompting and producing out of things like chat, gpt or Bard, but there are also images that can be created by a simple prompt. One of the open AI resources is Dolly, which creates images based on just word prompts or similar. If you upload an image and ask it to curtail it in some way, it can do that.
Speaker 1:It is completely crazy, as we've been, you know, talking about this and sitting on the committee that you chair, just watching how it's changed even in the past couple of months. But you mentioned open AI. Do you want to explain a little bit what we mean when we say open AI in the short?
Speaker 2:So open AI is a company and it is the company that has developed chat GPT. Bard is the offering by Google. Both have similar technology and of course they're sort of competitors in a way. And when we say open source generative AI, that's where it's a free product available to anyone who can log into a website and for the most part it's free. The open AI Dolly website actually does require purchasing credits to use it, play around with it, but so far the use of chat GPT remains free.
Speaker 2:The freedom creates concern of privacy. So you know, as a lawyer, we're always worried about maintaining the confidences of our clients and when you are looking at these new tools, this new technology, it's important to understand that you sort of get what you pay for. You wouldn't necessarily want to share in an open forum client information. You know if you're sitting in a coffee shop and you're having a conversation and people are listening to you and you're talking about a specific client and you name that client, that information is no longer protected by attorney client privilege. Think about ChatGPT as an open coffee shop. Any discussion you have in ChatGPT that identifies a client is then risking waiver of the attorney client privilege.
Speaker 1:Then whatever source you're using can learn from what you've told it. Is that correct? That's correct.
Speaker 2:One of the features of OpenAI's ChatGPT is that it's continuing to learn and it's learning from the data that you enter. There are non-open source generative AI tools that are still being developed, are developed are on the market and in those you typically will see restraints on using the data that you enter to train the model. That's a really important piece of contract negotiations with your technology vendors as attorneys, to make sure that it is clear that representation, that no information entered into the solution will be used to train the model.
Speaker 1:You don't want the other side learning from what you've put in. There's all kinds of issues that could come up.
Speaker 2:Yes, it's the other side. If you enter your client information and it's training the model and opposing counsel is looking for similar information on your company, it doesn't take much to run that search because the processing speed is really phenomenal and impressive. While ChatGPT still isn't necessarily up to date with the data, it has still collecting the information and we don't know when ChatGPT5 will be rolled out and what data that will have collected.
Speaker 1:Exactly, just a side note, I've heard that the amount of electricity and power energy it takes to run these models, these different platforms, is pretty insane, because of how fast it runs.
Speaker 2:Yes, Think about this. It was a similar conversation we had Probably I think I'm aging myself maybe nearly a decade ago, when the buzzword was blockchain and Bitcoin, and how many? What is the computer processing power to locate the Bitcoin on the blockchain and it's similar super computing powers that are necessary to process the data?
Speaker 1:It's wild. In the legal field, a couple of the things happening are there are client needs and wants, and then how to integrate this into our practice as a professional. We're working on both of those simultaneously, aren't we?
Speaker 2:Well, yes, We've always needed to make sure that we are providing our clients with the most cost-effective legal counsel possible utilizing tools and technology that we understand. That's the framework that attorneys should be confined to when we talk about generative AI tools understanding the requirements of confidentiality, the ethical obligations of protecting that confidential information, as well as the ethical obligations to make sure that whatever you produce as work product is as accurate to your knowledge as possible. That raises concerns on another pathway of open-source AI, which is hallucinations. It's not just open-source AI, by the way, it is all generative AI tools have the risk of hallucinating a response, because that artificial intelligence piece does not want to necessarily admit that it doesn't know the answer, or it might make an assumption to create an answer that sounds plausible but is inaccurate.
Speaker 1:That is completely wild right. It almost has an ego.
Speaker 2:It does. It's interesting because there's not necessarily other evidence of human emotion in the responses. But that's fascinating. It is so well-trained that it has to find an answer.
Speaker 1:I believe that's called a fixed mindset. If we were doing it, it doesn't want to be rushed. Some of the work that Bond is doing for itself, the work that you're leading, can help with the work we do for our clients. Can it Like policies use?
Speaker 2:Well, yes, I, along with a few other attorneys at the firm, are trained on AI governance. That provides us the baseline knowledge and the ability to guide organizations internally here, but then externally, in terms of what your artificial intelligence policy should say and whether you need to have a risk assessment performed to determine where are the uses of generative AI in your organization. What tools might you need to benefit your client or customer base, and that's also something that we are looking into. So, as you started the conversation a little earlier, kim, we need to balance the risks as well as the benefits for our clients and, in order to stay on top of this technology, we have been exploring different products that we would purchase and have a contractual relationship with, with all of the guarantees and warrants of any other technology support, as well as the due diligence and the auditing of the security measures taken by the companies that we rely upon.
Speaker 1:Yeah, so I know that's been something that the committee has been really focused on. But I think about this after you're talking. It's sort of like all the questions have to be answered. Is that who, what, when, where and how right? You have to figure out all of those within our ethical framework and then clients sort of have to do the same thing within their own industries.
Speaker 2:Yes, and I have. I've had a wide spectrum of reaction by clients and I know that I'm not alone in that, neither at bond nor you know sort of globally within law firms. To be honest is you have the one end of the spectrum, clients that are concerned that their attorneys and their outside council might be using generative tools that put their confidential information at risk To the other side of the spectrum, that say we will not engage a law firm if they're not going to optimize generative tools because we know that will be more cost effective. And so finding that safe and secure and ethical solution is a challenge for many law firms and I don't believe any one law firm as of today, in 2024, has all the answers, but I do think we're all working towards that goal.
Speaker 1:Yeah, absolutely, and that you touched on it briefly, but there's a variety of different types of training on the use of AI, on AI itself, that you and others at the firm have already done.
Speaker 2:Yeah. So there's so much to learn in this space. There's one, the governance, the overarching piece of it, and in that piece of it you're not necessarily just looking at the protection of confidential information, but you're looking at it even deeper, at the tools. Are they facilitating or propagating bias in their model? How do you dig deeply into an algorithm, a source, to understand what data is entered into this solution that will generate a response that is not bias? So that's one piece of the governance. It's the protection of confidential information, not just attorney client privilege. But we're talking, you know, cross industry confidential information, private information, as you know, when I've been on the podcast with my cybersecurity hat on, and we talk about protecting individual names, social security number, driver's license information. Are the tools protecting that information in a way such that, if there were a reach, would that data be exposed? So that's the governance piece of it. But then there's the practical side of training, and that's one of the trainings that I've completed and a couple of others at bond have is prompt engineering.
Speaker 2:The data you've received is only as good as the data you enter, so it's important to know how to craft the query in whatever solution you're using to facilitate the ability for that tool to create something to actually work.
Speaker 2:And that first response. By the way, unless you're asking for a poem that you've never otherwise written or you know lyrics to a song, if you're looking to use something to maybe draft an email to summarize a marketing plan, you're going to need to spend a lot of time looking at it and curtailing it to your needs, and so that's the other piece of it is. I've often been asked, since the launch of generative AI, you know, are lawyers going to be replaced? I do not see a scenario where lawyers will be replaced by generative AI. I do believe it will make us more efficient, more cost effective, but there is a necessity for attorney eyes on the product that comes out of any of these generative AI tools, not just to check the accuracy of it, but to make sure that it explains the answer to a client in the way that you understand your client wants to.
Speaker 1:Yeah, absolutely. There's so many pieces of it and the lawyer's ability to even craft those prompts with all the legal knowledge that they bring to that, so it's a huge undertaking and I'm really glad that the firm has already really full in. Yeah, I agree, so, I guess, with clients, what types of inquiries are you seeing regarding generative AI?
Speaker 2:Inquiries as to whether Bond has allowed its attorneys to use chat, gpt inquiries and prohibitions against the use of it, questions about what tools would be beneficial, what could be, you know, what practice areas might benefit from it. That's something that we're looking at as well, because, you know, when you talk about the legal industry as you know, cam, as an attorney there are many different practice areas that new tools are coming out for. You know, patent prosecution, trust in the state's work, there's just a vast ability to assist an M&A due diligence efforts. Where are these tools and how can we identify them appropriately? How can we know them to be comfortable with the security surrounding them and the accuracy of the output? And then, you know, make that investment for our firm.
Speaker 1:It's just fascinating how quickly this is all moving and how much there is to sort of dive into With this technology. I mean speaking for other professions as well, but certainly for ours. It seems that we'll need to be in generative AI. Whatever form the closed platforms, it's got to be in the hands of licensed attorneys to work safely and properly within the law, doesn't it?
Speaker 2:It does, and I want to just go back to while we're looking at these tools, we want to make sure that it's the best investment for our firm and our firm's clients.
Speaker 1:Right, absolutely, because we always we want whatever tools we use to be as sort of to the level that we expect everything to be Exactly yeah, all right, so, while we have to be on top of innovation, so you're both moving with some care and some speed, is that sort of what's happening right now?
Speaker 2:We're quickly adapting in a safe manner.
Speaker 1:Yes, I love that. Quickly adapting in a safe manner. I like that. That's right. We have to be so. Anything else on artificial intelligence, generative AI, that you'd want to mention for today's conversation?
Speaker 2:Always keep learning about these tools and it's hard to know what sources to trust. You know, it's often that I'll see images of celebrities and I know that they're not actually legitimate images and it just begs the question of the news articles that are being propagated through social media channels. So just be aware that generative AI is out there and available and those open source resources that people are using might not be the most trustworthy, and certainly continue to learn about the proper governance of it and any other questions I've always reachable and to make sure that the firm is staying on top of this evolving technology and the proper uses of it. We are continuing our series of the AI summit. We held one in Buffalo in November of 2023. And we will be taking that on the road for a New York City audience on March 5th and a Long Island audience on March 6th. For more information, you can check our website out.
Speaker 1:Thanks for letting us know that, jessica. I think it's important and a great way for people to keep learning, because I think we all have to be continuously learning just as generative AI is. So I was looking at my outline from our last conversation and I literally wrote the note, always looking for excuses to bring Jessica back. So I'm glad that you're all in on AI because I think it's going to be changing a lot and we're going to have a lot of opportunities to bring you back. So I hope you'll come back again soon, anytime, kim, thank you. Thank you for tuning into this episode of Legally Bond. If you're listening and have any questions for me, want to hear from someone at the firm or have a suggestion for a future topic, please email us at LegallyBond at BSKcom. Also, don't forget to rate, review and subscribe to Legally Bond where our podcasts are downloaded. Until our next talk, be well.
Speaker 3:Bond, seneca and King has prepared this communication to present only general information. This is not intended as legal advice, nor should you consider it as such. You should not act or decline to act based upon the contents. While we try to make sure that the information is complete and accurate, laws can change quickly. You should always formally engage a lawyer of your choosing before taking actions which have legal consequences. For information about our communication, firm practice areas and attorneys, visit our website BSKcom. This is Attorney Advertising.